Before joining UBC, I obtained a Master of Science in Statistics from the Vienna University of Technology, advised by Prof. Peter Filzmoser.
My research agenda comprises methodological and computational aspects of robust estimation in high-dimensional problems as well as their application to Biomedical Sciences. I am working on statistical methods with reliable performance under presence of adverse contamination anywhere in the data.
For regression problems, for instance, I work on estimators which are resilient to outliers in the response but also to unusual values in the (potentially) explanatory variables. If not handled appropriately, unusual values in the explanatory variables can have a much more detrimental affect on the analysis than outliers in the response alone.
University of British Columbia
- Winter 2019/20 Term 2: STAT 305 – Introduction to Statistical Inference
- Cohen Freue GV, Kepplinger D*, Salibián-Barrera M, Smucler E. Robust elastic net estimators for variable selection and identification of proteomic biomarkers. Annals of Applied Statistics.2019;13(4). online pdf (* in alphabetical order)
- Kepplinger D, Takhar M, Sasaki M, Hollander Z, Smith D, McManus B, et al. PGCA: An algorithm to link protein groups created from MS/MS data. PLOS ONE. 2017;12(5). online
- Kepplinger D, Filzmoser P, Varmuza K. Variable selection with genetic algorithms using repeated cross- validation of PLS regression models as fitness measure. preprint
- Kepplinger D, Templ M, Upadhyaya S. Analysis of energy intensity in manufacturing industry using mixed-effects models. Energy. 2013;59:754 – 763. online
A complete list of publications, conference presentations, and other research experience can be found in my CV.
I am maintaining several stable R packages on CRAN and Bioconductor as well as several experimental software tools available on my GitHub and UBC GitLab pages.
Algorithms for non-smooth optimization
C++ template library, wrapped in an R package, providing modern and fast algorithms for optimizing non-smooth functions (e.g., L1 regularized objective functions).GitLab Repository
Penalized Elastic Net S/MM-Estimators of Regression
Robust penalized elastic net S- and MM-estimators for linear regression.View on CRAN
Robust Linear Regression with Compositional Covariates
Methods for robustly fitting regression models where the explanatory variables are compositional. Includes bootstrap methods for classical robust regression and compositional robust regression.View on CRAN
Link Protein Groups Created from MS/MS Data
Protein Group Code Algorithm (PGCA) is a computationally inexpensive algorithm to merge protein summaries from multiple experimental quantitative proteomics data.View on Bioconductor